Why Location-Based Machine Learning Is the Smart Path to Personalization
People are good at forgiving the capriciousness of other people. When our normally gregarious barista has a morning when he forgets to smile, he is easily forgiven — chalk it up to “one of those days,” we all have them. However, our tolerance and patience with quirky, unpredictable technology is non-existent. Which begs the question: with so many brilliant advances in machine learning and artificial intelligence, why are so many businesses so far behind in adopting the kind of technology that can create true personalization?
Hire your own engineer
Anyone who has used voice recognition software understands that all it takes in one misinterpreted command and a user’s annoyance quickly snowballs into expletives. But what most people don’t realize is that machine learning saved SIRI from becoming Apple’s version of Ishtar — a big budget bust. Despite the verbal abuse SIRI takes on a daily basis, she is a machine learning success story. In a recent eye-opening article on Backchannel, Steven Levy details how machine learning and artificial intelligence permeate nearly all of Apple’s products and services.
We can forgive our fellow humans for erring, but machines are expected to be divine… all the time. This unbalanced expectation is why it is so imperative that businesses adopt mobile marketing technology that gets smart about personalization and engages without annoying.
Most small and medium-sized business owners would be happy enough to hire a summer intern to take care of their tech needs, so the thought of implementing a mobile marketing initiative that incorporates machine learning may seem a bit daunting. While mobile engagement still sounds abstract to many, it is becoming necessary — the good news is you don’t need your own Apple egghead to do it. Other pioneers of machine learning like Google, Facebook and Microsoft have helped democratize the technology so smaller businesses can get in the game.
What we “Like” vs. what we prefer
Every business — from hotels to hospitals to the corner café — is at the mercy of reactionary negative reviews. But when that same reviewer is satisfied, their mounds of praise and proactive positive reviews tend to fall by the wayside. Even the data gathered from declared preferences such as social media “favorites” are misleading and a poor predictor of real spend. Just because somebody claims to like the latest Haruki Murakami novel doesn’t mean they actually finished it — or even bought it.
Because of the discrepancy between preference perception and preference reality, the data from traditional machine learning is non-representative. The algorithms operating on it rely on brute-force to try to make up for insufficient contextualization. The result is the equivalent of a Facebook post from the top of the Eiffel Tower, when in reality the user is at the bottom of their parents’ basement. Adding location aware technology to machine learning is the smart path to true personalization.
Good data makes a good neighborhood
The foundation of location-based machine learning takes as its foundation an opt-in, whereby users allow a mobile app to submit information about geographical locations visited throughout the day. Geo-data and dwell-time can be correlated against online rosters of locations with category keywords to make inferences about a user’s preferences, buying habits, and demographics. It can supplement declared interests, spend events, and all other types of online data to grow exponentially better profiles over time. The patterns it identifies are more meaningful and representative of what the user really prefers. Adding location to the equation replaces assumptions with accurate conclusions.
Think beyond the app and embrace the next generation of customer engagement
To be successful, mobile engagement must be precise and tactful. If the communications are annoying, embarrassing or inconvenient, the customer will quickly and unceremoniously ditch your app. A single download is hard enough — forget about second chances.
Adding location-based machine learning means filtering out those annoying promotional bad guesses and delivering useful content, offers and promotions. The app can predict real interest based on long-term, continuous, real-time sampling. Businesses that can capitalize on their customers interests at the right time, in the right place will establish their brand as valuable, personalized and trusted.
It’s what you do with the data
When data from location-based machine learning is aggregated from large sample sizes, it can give absolutely stellar reports. How often and in what locations do loyalty users visit the competition? What can a business infer based on their movements about age, sex, household income, and spending habits? With the right data a brand can keep the relationship with the customer warm while they’re off-property, out-of-town or on a budget. Offering the right deals and communications that are relevant to real interests means that the relationship between man and machine is getting better all the time. The big guys like Apple, Amazon Google and Microsoft have laid the foundation, now it’s time for smaller businesses to build their own castles in the sky.
Michael Garvin is the CEO at RoamingAround, an industry and thought leader for location-based services with particular expertise in Hospitality. The RoamingAround platform is the most sophisticated, most flexible, most comprehensive solution currently on the market for hyper-personalized, location-contexted, dynamic mobile engagement and machine learning—and one of the very few to focus on the particular opportunities offered in Hospitality.